prognostics and health management software for gas turbine

Proceedings of GT2007 

ASME Turbo Expo 2007: Power for Land, Sea, and Air 

May 14-17, 2007, Montreal, Canada 

GT2007-27984 

PROGNOSTICS AND HEALTH MANAGEMENT SOFTWARE FOR GAS TURBINE ENGINE BEARINGS 

Michael J. Roemer and Carl S. Byington 

Impact Technologies Rochester, NY 14623, USA 

mike.roemer@impact-tek.com 

ABSTRACT 

Based on the results of a successful Phase I and II SBIR 

program performed by Impact Technologies, a suite of 

Prognostics and Health Management (PHM) algorithms have 

been developed for detecting incipient faults in the critical 

bearings associated with aircraft gas turbine engines. The 

component-level prognostic approach is presented that utilizes 

available sensor information from vibration transducers, along 

with material-level component fatigue models to calculate 

remaining useful life for the engine’s critical components. 

Specifically, correlation between the sensed data and fatiguebased 

damage accumulation models were developed to provide 

remaining useful life assessments for life limited components. 

The combination of health monitoring data and model-based 

techniques provides a unique and knowledge rich capability 

that can be utilized throughout the bearings’s entire life, using 

model-based estimates when no diagnostic indicators are 

present and using the monitored vibration features at later 

stages when incipient failure indications are detectable, thus 

reducing the uncertainty in model-based predictions. A 

description and specific implementation of this prognosis 

approach with application to high speed bearings is illustrated 

herein, using gas turbine engine and bearing test rig data as 

validation for the methods. 

INTRODUCTION 

An integrated prognostic capability can be achieved for 

gas turbine engine components such as rolling element bearings 

and gears through the fusion of health state awareness data and 

model-based damage prediction. The technical approach is 

based on event driven integration of diagnostic features and 

physics-based modeling. Since bearings and gears have many 

failure modes and there are many influencing factors, a 

modular approach is taken in the design. In addition, due to the 

many potential failure modes that exist for engine drivetrain 

components, complete discussion all such modes and their 

failure progression is beyond the scope of this paper. 

Therefore, this paper will specifically focus on the rolling 

element contact fatigue failure mode referred to as spalling. 

However, other failure modes may cause conditions that result 

in spalling (3), which are also addressed. For instance 

scratches, corrosion pits or manufacturing defects produce 

localized stress concentrations that prematurely cause spalling. 

Some failure modes are not well suited to the concepts of 

prognosis. For instance, maintenance induced failures 

(misalignment, incorrect installation, etc.) are impossible to 

predict, but result in conditions that can be tracked and trended. 

Contamination of the lubricant is another very common failure 

mode, but is often impossible to predict when such a condition 

will become prevalent. However, it is possible to detect the 

conditions, through sensor information, that precede many of 

these failures and adjust the prediction accordingly, such as 

detection of water in the lubricant would reduce the fatigue life 

prediction. 

This paper’s approach to engine component fault detection and 

prediction, with specific application to high speed bearings, 

entails the three functional blocks as shown in Figure 1: Sensed 

Data, Current Bearing Health, and Future Bearing Health. The 

fault detection and prediction process begins with the Sensed 

Data module. Signals indicative of bearing health (vibration, 

acoustic emissions, temperature, etc.) are monitored to 

determine the current bearing condition. Diagnostic features 

extracted from these signals are then passed on to a Current 

Bearing Health module. These diagnostic features are low 

bandwidth, processed features, such as root mean square 

(RMS), kurtosis, and high frequency enveloped features, to be 

discussed in subsequent sections. In addition, engine speed and 

maneuver induced loading can be calculated and used as inputs 

to bearing health models. Also extracted are characteristic 

features that can be used to identify failure of a particular 

bearing component (ball, cage, inner or outer raceway). 

1 Copyright © 2007 by ASME

•Engine speed 

•Maneuvers 

•Vibration 

•Oil pressure 

•Oil temperature 

•Oil quality 

Mission 

Profile 

Sensed Data 

Future Bearing Health 

Spall Initiation Model 

x 

+ 

y 

+ 

z 

3 

⎛ 

Q 

c 

⎞ 

L 

= 

⎜ 

⎟ 

⎝ 

Q 

⎠ 

Yu-Harris 

No 

Current Bearing Health 


⎛ 

Q 

⎞ 

L 

⎝ 

Q 

⎠ 

x 

+ 

y 

+ 

z 

3 

c 

= ⎜ 

⎟ 

Yu-Harris 

Spall? 

Fusion 

engine rotors, blade passing, and gear mesh in a running 

engine. This difficulty is illustrated in Figure 2. 

RUL 

Progression Model 

⎛ 

⎛ 1 

⎞ 

e 

c 

⎜ 

⎟ 

∝ 

N 

⎜ 

∫ τ 

d 

v 

ln 

⎝ 

S 

⎠ 

⎝ 

e 

⎞ 

v ⎟ 

⎠ 

Yes 

Figure 1. Overall Bearing Prognostic Architecture 

Central to the Current Bearing Health step is a material-level, 

rolling contact fatigue (RCF) model. This model utilizes 

information from the Sensed Data module to calculate the 

cumulative damage sustained by the bearing since it was first 

installed. Life limiting parameters used by the RCF model 

such as load and lubricant film thickness are derived from the 

sensed data using physics-based and empirical models. 

Utilizing an information fusion process, this probability is 

combined with the extracted features that are indicative of 

spalling. Combining the model output with the features 

improves the robustness and accuracy of the prediction. 

Whether or not a spall is detected determines the next step of 

estimating Future Bearing Health. If a spall does not currently 

exist the spall initiation prognostic module is used to forecast 

the time to spall initiation. This forecast is based on the same 

model that is used to assess the current probability of spall 

initiation, but instead the model uses projected future operating 

conditions (loads, speeds, etc.) rather than the current 

conditions. Then the initiation results are passed to the 

progression model, which also uses the mission profile to allow 

an accurate prediction on the time from spall initiation to 

failure. If a spall currently exists, the spall initiation prognostic 

module is bypassed and only the spall progression model is 

used. 

HIGH FREQUENCY VIBRATION INDICATORS 

Development of advanced vibration feature analysis that 

can detect early spalling characteristics is a critical step in the 

design of the integrated fault detection and prognosis 

architecture mentioned above. Although bearing characteristic 

frequencies are easily calculated, they are not always easily 

detected by conventional frequency domain analysis 

techniques. Vibration amplitudes at these frequencies due to 

incipient faults (and sometimes more developed faults) are 

often indistinguishable from background noise or obscured by 

much higher amplitude vibration from other sources including 

Figure 2. Incipient Bearing Fault Signatures Difficult to Extract 

Vibro-acoustic data sources provide some of the most reliable 

quantitative indicators of bearing, gear, and rotating component 

fatigue that is available (1). These indicators are typically 

spread throughout the vibro-acoustic regime. The vibration 

monitoring software developed as part of the SBIR Phase I/II 

program discussed herein applies advanced high frequency 

incipient fault detection and diagnostic algorithms on high 

frequency vibration monitoring sensor data collected from 

bearings and accessory gearboxes in gas turbine engines. 

Monitoring the high frequency range of the spectrum for 

increasing levels of vibrations is an effective method of 

identifying and tracking bearing and rolling element condition 

(2,3). Early material distress and incipient faults are most 

detectable at higher frequencies and thus an indication at this 

point will provide the greatest detection horizon. 

Vibro-acoustic data could be derived from multiple sensing 

systems to increase confidence of bearing fault prediction. 

Specific fault frequencies are clearly identifiable in the vibroacoustic 

regime of 10 through 100 kHz, using demodulation or 

enveloping techniques for bearings. The conceptual basis of 

the early fault detection is shown in Figure 3. As wear 

increases the noise drops in frequency range. Impact has 

developed its own software modules referred to as 

ImpactEnergy to evaluate these frequencies. During the 

incipient failure stage 2, slight defects begin to ring the bearing 

at natural frequencies (f n ) and sidebands appear around f n . At 

failure Stage 3, bearing defect frequencies and harmonics 

appear if the overall machinery noise is not too high. As wear 

progresses more harmonics appear with stronger sidebands 

around defect frequencies and f n . Wear is now visible. High 

frequency demodulation and enveloping can be fused to 

confirm Stage 3 progression of damage. At the very end of life, 


the magnitudes of 1X RPM are affected and more harmonics 

appear. 

The generation of vibro-acoustic indicators is best understood 

with a simple example of the failure progression and 

symptoms. A fault in a mechanical component such as a rolling 

element will create an impulse every time the defect rolls over 

the defective area of the race. This will cause the natural 

frequencies of the bearing elements and the housing structure 

to be excited. The result will be an increase in the vibration 

energy at these high frequencies, usually greater than 10 kHz. 

Vibration data to monitor this process may be collected from 

accelerometers, laser interferometer and acoustic emission 

(AE) sensors. The high frequency data collected using contact 

and non-contact sensors are typically too large to handle for 

fault detection and important features must be extracted from 

the sea of data. The high bandwidth data is processed in an 

advanced multi spectral feature extraction module. This system 

extracts time and frequency domain features of the broad and 

narrow band data. These features can be used directly as inputs 

to the fault classifier module. Trending and baseline vs. actual 

comparisons are two common techniques that form the basis 

for classification. It uses a combination of frequency domain 

and time domain feature extraction techniques that are briefly 

discussed in this section (4,5,6). 

feature extraction technique (5), which extracts energy from 

highly sensitive regions in the frequency domain similar to the 

general demodulation process illustrated in Figure 4. The basic 

enveloping process consists of first band pass filtering of the 

raw vibration signal. Second, the band pass filtered signal is 

full waved rectified to extract the envelope. Third, the rectified 

signal is passed through a low pass filter to remove the high 

frequency carrier signal. Finally, the signal has any DC content 

removed. 

The ImpactEnergy algorithms offer a distinct advantage over 

the vibration monitoring techniques that use time domain 

features like Root Mean Square (RMS) and frequency domain 

features from traditional FFT peak picking methods. This is 

because the vibration signatures from bearing degradation often 

consist of impact events that are characterized by high 

frequency, short-duration bursts of energy. With normal Fast 

Fourier Transform (FFT) analysis, these impact events translate 

to the frequency domain as small harmonic amplitudes 

distributed over a broad frequency range that are easily buried 

by machine noise. Similarly RMS and Kurtosis are not 

significantly affected by such short burst of high frequency low 

intensity vibrations. The demodulation or enveloping (3, 5) 

process was therefore developed to detect impulse events much 

easier than traditional analysis techniques allow. In short, 

enveloping differentiates between the broadband energy due to 

failure effects and the energy due to normal system noise. A 

high frequency carrier demodulation technique is applied by 

using a bandpass filter that is centered on the expected carriers. 

The bands used for the carriers are identified based on Impacts 

proprietary knowledge of vibration monitoring. The technique 

used for isolating these bands is similar to the AM radio tuning 

technique. 

Figure 4. High Frequency Demodulation Process 

Figure 3. Frequency Ranges for Bearing Fault Detection 

The specific incipient fault detection algorithm called 

ImpactEnergy is a multi-band, enveloping-based vibration 

ENGINE BEARING RIG SEEDED FAULT TESTING 

Under the SBIR program, the authors have designed and 

demonstrated software that can detect spall initiation and assist 

an aircraft engine manufacturer in the characterization of spall 

propagation behavior of high speed main shaft ball bearings 

used in turbofan engines. The angular contact ball bearings that 

were tested were operated at simulated engine conditions in a 


two-bearing test rig located at the bearing manufacturer’s 

location. The authors assembled and installed a portable test 

cell monitoring system composed of eight vibration sensors, 3 

data acquisition cards (14 channels total) with signal 

conditioning, data processing controller and storage and 

retrieval hardware and software. The authors assessed the 

preferred sensor location by performing an experimental modal 

analysis (or “ping” test) during an on-site visit prior to testing. 

As a result four accelerometers were installed on the outer race 

casement for the test and slave bearing (two in axial and two in 

radial vertical position), two accelerometers were installed on 

the exterior housing one radial vertical and one radial 

horizontal and the remaining two sensors were placed at the 

output of the speed increaser. The data collection frequency 

was 204,800 Ks/s. 

Test 6 was another test with spall progression and 

detection as the goal. The results of the outer race feature for 

the three high information bands are shown in Figure 6. Test 6 

was characterized by frequent starts-stops and inspections 

resulting in variations in the feature values. However, the 

general trend in Bands 2 and 3 reflects an outer race fault 

progression with a quantum jump in the feature value around 

file 2000 (4000 minutes or 67 hours) representing a spurt in the 

spall size and this feature growth and variability in the feature 

persisted till the test was stopped due to excessive noise and 

broadband vibration around 120 hours. 

Fourteen tests were performed including baseline operation and 

accelerated mission fault progression tests. Throughout the 

testing, periodic disassembly of the test rig was conducted to 

assess spall propagation of the bearing race for the purpose of 

system validation. Cumulative run times for each test vary 

based on operating speed and load conditions. Impact 

monitored and processed data from all available tests and the 

results from two of the tests are summarized below. Each test 

began with an induced outer race incipient fault and allowed 

the fault to progress under constant operating conditions. This 

section contains results of two typical tests. 

Spall Size: 

0.17”x0.17” 

Spall Size: 

0.56”x0.34” 

Test 3 Results 

Test 3 had an approximate duration of 115 hours, during 

which an outer race fault progressed to failure. Figure 5 shows 

the outer race fault features for three high information bands 

selected by ImpactEnergy TM . Bands 2 and 3 show good 

reaction to fault progression. Each data file represents 2 

minutes of data, for the continuous data collection process. 

Spall Size: 

0.22”x0.28” 

Spall Size: 

0.40”x0.40” 

Spall Size: 

0.58”x0.43” 

Spall Size: 

0.90”x0.50” 

Figure 6. ImpactEnergy Outer Race Feature Trends from Test 6 


The final set of results to be presented within this paper 

is from Test 12, which was a test to examine the stability and 

robustness of various vibration features during transient engine 

operation. This test began with an indent placed on the outer 

race of the bearing to allow for relatively quick initiation of the 

spall to occur. The results of the outer race feature for a 

selected information band is shown in Figure 6. Test 12 was 

characterized by continuous cycling of the speed and load on 

the bearing. As expected, the features extracted from the 

multiple ImpactEnergy bands were correlated with the speed 

and load fluctuations. These results depict a need for a 

smoothing algorithm to be used in combination with a mode 

detection process, if the resulting feature is to be used for 

fault/failure tracking for prognosis purposes. 

Figure 5. ImpactEnergy Outer Race Feature Trends from Test 3 



prediction by updating the model to reflect the fact that fault 

initiation has occurred. Subsequent predictions of the 

remaining useful component life can then be more weighted on 

fault progression rather than initiation models. 

Figure 7. ImpactEnergy Outer Race Feature Trends from Test 

12 

The analysis and results of Tests 3, 6 and 12 are typical of all 

14 tests. The results show that ImpactEnergy can 

consistently detect fault initiation and progression in bearings 

in a test cell, while mounting sensors on the outer casings of 

the engine. The outer casing mount is relevant for engine 

environments, because the transducer locations near to the 

bearings are not conducive to sensor survivability. The fault 

detection using outer casing as the mounting point for the 

accelerometers also shows that ImpacEnergy is capable of 

exploiting the transmissibility of the rig to provide good health 

state seperability and detect fault progression. The test results 

also show that start up and shut downs lead to cyclic variations 

in the feature values. However, despite these variations, the 

confidence level of incipient and severe fault detection is high. 

MATERIAL-LEVEL SPALL INITIATION MODELING 

Material-level prognosis models and sensor-based 

diagnostic approaches offer complementary condition 

assessment information that can be fused together to achieve a 

comprehensive diagnostic/prognostic capability throughout a 

components life. Model-based approaches provide valuable 

damage accumulation information on critical components well 

in advance of failure indications. However, due to modeling 

uncertainties, these long-range predictions typically have broad 

confidence bounds. Sensor-based approaches, on the other 

hand, can provide direct or indirect measures of component 

condition that can be used to update the modeling assumptions 

and reduce the uncertainty in the RUL predictions. 

To achieve a comprehensive diagnostic/prognostic capability 

throughout the life of critical engine components, model-based 

information can be used to predict the initiation of a fault. In 

the best scenario, these predictions can prompt “just in time” 

maintenance actions to prevent the fault from developing. 

However, due to modeling uncertainties, incipient faults may 

occasionally develop earlier than predicted. In these situations, 

sensor-based diagnostics complement the model-based 


A variety of theories exist for predicting spall initiation 

from bearing dimensions, loads, lubricant quality, and a few 

empirical constants. Many modern theories are based on the 

Lundberg-Palmgren (L-P) model that was developed in the 

1940’s (1). A model proposed by Ioannides and Harris (I-H) 

improved on the L-P model by accounting for the evidence of 

fatigue limits for bearings (6). Yu and Harris (Y-H) proposed a 

stress-based theory in which relatively simple equations are 

used to determine the fatigue life purely from the induced stress 

(7). This approach depends to a lesser extent on empirical 

constants, and the remaining constants may be obtained from 

elemental testing rather than complete bearing testing as 

required by L-P. 

The fundamental equation of the Y-H model stated in Equation 

1 relates the survival rate (S) of the bearing to a stress weighted 

volume integral as shown. The model utilizes a material 

property for the stress exponent (c) to represent the material 

fatigue strength, and the conventional Weibull slope parameter 

to account for dispersion in the number of cycles (N). The 

fatigue initiating stress (τ) may be expressed using Sines’s 

multi-axial fatigue criterion for combined alternating and mean 

stresses, or as a simple Hertz stress (8). 

⎛ 1 ⎞ 

e c 

ln⎜ 

⎟ ∝ N ⎜ 

⎝ ⎠ 

∫τ 

d 

S 

v 

⎛ 

⎝ 

⎞ 

v⎟ 

⎠ 

For simple Hertzian stress, a power law is used to express the 

stress-weighted volume. In Equation 2, below, λ is the 

circumference of the contact surface, and a and b are the major 

and minor axes of the contact surface ellipse. The exponent 

values were determined by Yu and Harris for b/a ≈0.1 to be 

x=0.65, y=0.65, and z=10.61. Yu and Harris assume that these 

values are independent of the bearing material. 

e 

(1) 

c 

x y z 

∫ 

τ dA ⋅ λ ∝ a b τ λ 

(2) 

A 

According to the Y-H model, the life (L 10 ) of a bearing is a 

function of the basic dynamic capacity (Q c ) and the applied 

load (Q) as stated below in Equation 3. Where, the basic 

dynamic capacity is given in Equation 4. A lubrication effect 

factor may be introduced to account for variations in film 

thickness due to temperature, viscosity, and pressure. 

Although this approach was developed for angular contact ball 

bearings, it is extendable to other bearing types such as tapered 

roller and cylindrical bearings. 


Q 

L 

C 

10 

⎛ Q 

= ⎜ 

⎝ 

c 

Q 

⎞ 

⎟ 

⎠ 

= A ΦD 

1 

x+ 

y+ 

z 

3 

( 2z−x− 

y ) 

( z+ 

x+ 

y) 

(3) 

(4) 

TABLE 1. ROLLING CONTACT FATIGUE 

TESTER DIMENSIONS (MM) 

Rod diameter (Dr) 9.52 

Ball diameter (Db) 12.70 

Pitch diameter (Dm) 22.23 

⎡ 

⎛ 

Φ = ⎢ 

⎢ 

⎜ 

⎣ 

⎝ 

T 

T 

1 

⎞ 

⎟ 

⎠ 

z 

u 

( ) ( 2z−x− 

y 

DΣρ 

) 

3 

* z−x 

* 

( a ) ( b ) 

z− 

y 

d 

D 

⎤ 

⎥ 

⎥ 

⎦ 

−3 

z+ 

x+ 

y 

(5) 

Where: 

A 1 = Material property 

T = A function of the contact surface dimensions 

T 1 = value of T when a/b = 1 

u = number of stress cycles per revolution 

D = Ball diameter 

ρ = Curvature (inverse radii of component) 

d = Component (race way) diameter 

a*= Function of contact ellipse dimensions 

b*= Function of contact ellipse dimensions 

MODEL VALIDATION 

An updated version of this model was developed by the 

authors with a stochastic “wrapper” used to account for 

variability in the model parameters (9,10). Validation of a 

stochastic spall initiation model requires a statistical 

comparison of actual fatigue life values to predicted model 

values. Acquiring sufficient numbers of actual values is not a 

trivial task. Under normal conditions, it is not uncommon for a 

bearing life value to extend past 100 million cycles, prohibiting 

normal run-to-failure testing. 

Accelerated life testing is one method used to rapidly generate 

many bearing failures. By subjecting a bearing to high speed, 

load, and/or temperature, rapid failure can be induced. There 

are many test apparatus used for accelerated life testing 

including ball and rod type test rigs. One such test rig is 

operated by UES, Inc for AFRL at Wright Patterson Air Force 

Base in Dayton, OH. A photo of the test rig is shown in Figure 

8. This rig consists of three 12.7 mm diameter balls contacting 

a 9.5 mm rotating central rod; see Table 1 for exact 

dimensions. The three radially loaded balls are pressed against 

the central rotating rod by two tapered bearing races that are 

thrust loaded by three compressive springs.. Notice two 

accelerometers mounted on the top of the unit. The larger 

accelerometer is used to automatically shutdown the test when 

a threshold vibration level is reached, the other measures 

vibration data for analysis. 

Figure 8. Rolling Contact Fatigue Tester 

By design, the rod is subjected to high contact stresses. Due to 

the geometry of the test device, the 222 N (50 lbs) load applied 

by the springs translates to a 942 N (211 lbs) load per ball on 

the center rod. Assuming Hertzian contact for balls and rod 

made of M50 bearing steel, the 942 N radial load results in a 

maximum stress of approximately 4.8 GPa (696 ksi). This 

extremely high stress causes rapid fatigue of the bearing 

components and can initiate a spall in less than 100 hours, 

depending on test conditions including lubrication, 

temperature, etc. Since failures occur relatively quickly, it is 

possible to generate statistically significant numbers of events 

in a timely manner. 

For validation purposes M50 rods and balls were tested at 

room temperature (23°C). The results of these tests are shown 

in Table 2. A summary plot of fatigue life for 30 tests is 

shown in Figure 9. 

TABLE 2. RCF FATIGUE LIFE RESULTS 

Failures (#) Susp (#) Susp Time 

(cylces) 

29 1 (ball failed) 83.33 


The median ranks of the actual lives are plotted against the 

cumulative distribution function (CDF) of the predicted lives. 

The model predicted lives are slightly more conservative (in 

the sense that the predicted life is shorter than the observed 

life) once the cumulative probability of failure exceeds 70%. 

However since bearings are a critical component, the main 

interest is in the left most region of the distribution where the 

first failures occur and the model correlates better. 

Figure 9. RCF Fatigue Life Results 

As stated above, one of the issues with empirical/physics based 

models is their inherent uncertainty. Assumptions and 

simplifications are made in all modeling and not all of the 

model variables are exactly known. Often stochastic 

techniques are used to account for the implicit uncertainty in a 

model’s results. Statistical methods are used to generate 

numerous possible values for each input. 

A Monte Carlo simulation was utilized in the calculation of the 

bearing life distribution. Inputs to the model were represented 

by normal or lognormal distributions to approximate the 

uncertainty of the input values. The developed stochastic Y-H 

type model was used to simulate the room temperature M50 

RCF tests. Figure 10 shows the results for a series of the room 

temperature RCF tests on the M50 bearing material. This test 

was run at 3600 RPM with the 7808K lubricant. The plot 

shows the number of central rod failures versus cycles to 

failure. The predicted life from the model is superimposed on 

the actual test results. This predicted distribution shown in red 

was calculated from the model using one million Monte Carlo 

points. 

Calculation of median ranks is a standard statistical procedure 

for plotting failure data. During run-to-failure testing there are 

often tests that either are prematurely stopped before failure or 

a failure occursof a component other than the test specimen. 

Although the data generated during these failures are the mode 

of interest, they provide a lower bound on the fatigue lives due 

to the failure mode of interest. One method for including this 

data is by median ranking. 

The median rank was determined using Benard's Median 

Ranking method, which is stated in Equation 6 below. This 

method accounts for tests that did not end in the failure mode 

of interest (suspensions). In the case of the ball and rod RCF 

test rig, the failure mode of interest is creation of a spall on the 

inner rod. The time to suspension provides a lower bound for 

the life of the test article (under the failure mode of interest), 

which can be used in reliability calculations. During the testing 

on the RCF test rig, significant portions of the tests were 

terminated without failure after reaching ten times the L 10 life. 

There were also several tests that ended due to development of 

a spall on one of the balls rather than on the central rod. 

Where: 

Benard's Median Rank 

= 

AR = Adjusted Rank 

( AR − 0.3) 

( N + 0.4) 

N = Number of Suspensions and Failures 

(6) 

The adjusted rank is calculated below. 

AR = 

( ReverseRank) x ( PreviousAdjusted Rank) + ( N + 1) 

ReverseRank + 1 

(7) 

Although the test does not simulate an actual bearing assembly 

in an engine, it does simulate similar conditions (1). Materials 

and the geometry of the bearing and the lubricants are the same 

for the test rig as they are in the T-63 engine. The test rig 

results validate the model's ability to predict the fatigue life of 

the material under similar conditions to an operating engine. 

Figure 10. Room Temp Results vs. Predicted 


CONCLUSIONS 

To achieve a comprehensive fault detection and prediction 

capability throughout the life of critical engine components 

such as bearings, a technical approach that implements a 

combination of advanced vibration fault detection and materiallevel 

modeling has been described. In the best scenario, such 

models can be used to prompt “just in time” maintenance 

actions to prevent the fault from developing. However, due to 

modeling uncertainties, incipient faults may develop earlier 

than predicted or without warning. In these situations, sensorbased 

(or feature-based) diagnostics complement the modelbased 

prediction by updating the model to reflect the fact that 

fault conditions may have occurred. These measurement-based 

approaches can provide direct correlation of the component 

condition that can be used to update the modeling assumptions 

and reduce the uncertainty in the RUL predictions. 

8. Sines, and Ohgi, “Fatigue Criteria Under Combined 

Stresses or Strains”, ASME Journal of Eng. Materials and 

Tech., Vol. 103, pp. 82-90, 1981. 

9. Abernethy, Robert B., "The New Weibull Handbook," 

1998. 

10. Kacprzynski, G., et al., “Calibration of Failure 

Mechanism-Based Prognosis with Vibratory State 

Awareness Applied to the H-60 Gearbox” Proceedings of 

the 2003 IEEE Aerospace Conf., Big Sky, MT. 

11. Roemer, M., and Kacprzynski, G., “Development of 

Diagnostic and Prognostic Technologies for Aerospace 

Health Management Applications” 2001 IEEE Aerospace 

Conf., Big Sky, MT. 

These advanced prognostic/diagnostic algorithms utilize a 

fusion architecture that optimally combines sensor data, with 

probabilistic component models to achieve the best decisions 

on the overall health of oil-wetted components. By utilizing a 

combination of health monitoring data and model-based 

techniques, a comprehensive component prognostic capability 

can be achieved throughout a components life, using modelbased 

estimates when no diagnostic indicators are present and 

monitored features such as vibration condition indicators at 

later stages when failure indications are detectable. 

REFERENCES 

1. Toth, Douglas K., Saba, Costandy S., Klenke, Christopher 

J. "Minisimulator for Evaluating High -Temperature 

Candidate Lubricants Part I - Method Development," 

University of Dayton Research Institute - Aero Propulsion 

Directorate USAF, Dayton, OH, 2001. 

2. Glover, Douglas. "A Ball-Rod Rolling Contact Fatigue 

Tester," Rolling Contact Fatigue Testing of Bearing Steels, 

ASTM STP 771, ASTM, 1982, pp. 107-125. 

3. Harris, T. (4 th Edition 2001), Rolling Bearing Analysis, 

John Wiley & Sons, New York. 

4. Wensig, J. A.," On the Dynamics of Ball Bearings," PhD 

Thesis, University of Twente, The Netherlands, pp. 90, 

1998. 

5. Orsagh, Rolf F., Sheldon, Jeremy, Klenke, Christopher J., 

"Prognostics/Diagnostics for Gas Turbine Engine 

Bearings," Presented at STLE Annual Meeting 2003, 

STLE, NY, NY, publication pending. 

6. Ioannides, and Harris, “A New Fatigue Life Model for 

Rolling Bearings”, Journal of Tribology, Vol. 107, pp. 

367-378, 1985. 

7. Yu, and Harris, “A New Stress-Based Fatigue Life Model 

for Ball Bearings”, Tribology Transactions, Vol. 44, pp. 

11-18, 2001.

prognostics and health management software for gas turbine

Create successful ePaper yourself

Delete template?

Save as template?